If $\alpha=0$ then $G_Y=1$, so $Y$ is a constant variable of value $0$. If $\alpha\ne0$,$$G_Y=\frac{1}{1-\alpha}\frac{1}{1+\frac{\alpha}{1-\alpha}s}=\sum_{n\ge0}\frac{(-\alpha)^n}{(1-\alpha)^{n+1}}s^n$$isn't the PGF of any distribution due to its including negative coefficients.
If $\alpha=0$ then $G_Y=1$, so $Y$ is a constant variable of value $0$. If $\alpha\ne0$,$$G_Y=\frac{1}{1-\alpha}\frac{1}{1+\frac{\alpha}{1-\alpha}s}=\sum_{n\ge0}\frac{(-\alpha)^n}{(1-\alpha)^{n+1}}s^n$$isn't the PGF of any distribution due to its including negative coefficients.